From petro at cpetro.us Mon Nov 4 18:18:43 2013 From: petro at cpetro.us (Christopher Petro) Date: Mon, 4 Nov 2013 11:18:43 -0700 Subject: udev rules biosdevnames. Message-ID: I've got an odd one here, and I'm not sure where to start. We're installing RHEL 6.3 and doing bonding of the first (or zeroth) onboard and the first offboard. To do this consistently we really would prefer to use the biosdevnames functionality (e.g. em1 bonds with p2p1). (We're using Cobbler to provision these systems, if that matters). This worked for the first 6 machines I kickstarted, but either I'm holding my jaw different, or something has changed in numbers 7 and 8. The first round works fine--the OS installs using em1, em2 and p2p1, the appropriate schtuff gets created in /etc/sysconfig/network-scripts (meaning ifcfg-bond0, ifcfg-em1 ifcfg-em2 and ifcfg-p2p1. But /etc/udev/rule.d/70-persistent-net-rules gets created with eth0 through eth5, which prevents the ifcfg'd interfaces from coming up. I put a rm /etc/udev/rule.d/70-persistent-net-rules in kickstart, and it didn't help. After the first reboot I tried deleting that file and it was recreated on the next reboot. 70-persistent-net.rules references /lib/udev/write_net_rules, which SEEMS to be reading /etc/.../network-scripts/ifcfg* to do its stuff. When I edit the 70-persistent-net.rules file I get my networking back, but the whole point of kickstart is that I shouldn't have to do that, right? -------------- next part -------------- An HTML attachment was scrubbed... URL: From atodorov at redhat.com Tue Nov 12 14:48:16 2013 From: atodorov at redhat.com (Alexander Todorov) Date: Tue, 12 Nov 2013 16:48:16 +0200 Subject: pykickstart usage of OrderedDict and SNAKE Message-ID: <52823FB0.80007@redhat.com> Hi guys, I'm working together with Fedora QA to configure a Beaker lab and a SNAKE server in order to be able to run installation test jobs (among others) for Fedora. I'm also opening up some of the test cases previously used internally at Red Hat. Currently we're seeing the following issues: 1) SNAKE templates depend on pykickstart and need a recent version for the RHEL7 constant and support for latest syntax 2) The service is running on a RHEL 6 box with Python 2.6. After rebuilding pykickstart-1.99.45-1 from SRPM and upgrading I hit this: > Traceback (most recent call last): > File "/usr/sbin/snake-server", line 117, in _dispatch > return DocXMLRPCServer._dispatch(self, method, params) > File "/usr/lib64/python2.6/SimpleXMLRPCServer.py", line 418, in _dispatch > return func(*params) > File "/usr/lib/python2.6/site-packages/snake/ksdb.py", line 167, in generate > k = PyKickstart(version=ksversion) > File "/usr/lib/python2.6/site-packages/snake/kickstart.py", line 61, in __init__ > self.setversion(version) > File "/usr/lib/python2.6/site-packages/snake/kickstart.py", line 171, in setversion > Kickstart.setversion(self,version) > File "/usr/lib/python2.6/site-packages/snake/kickstart.py", line 77, in setversion > self.handler = handlerclass() > File "/usr/lib/python2.6/site-packages/pykickstart/base.py", line 275, in __init__ > self._registerCommands(mapping, dataMapping, commandUpdates, dataUpdates) > File "/usr/lib/python2.6/site-packages/pykickstart/base.py", line 348, in _registerCommands > from pykickstart.handlers.control import commandMap > File "/usr/lib/python2.6/site-packages/pykickstart/handlers/control.py", line 21, in > from pykickstart.commands import * > File "/usr/lib/python2.6/site-packages/pykickstart/commands/__init__.py", line 20, in > import authconfig, autopart, autostep, bootloader, btrfs, clearpart, cdrom, device > File "/usr/lib/python2.6/site-packages/pykickstart/commands/autopart.py", line 24, in > from collections import OrderedDict > ImportError: cannot import name OrderedDict From what I see OrderedDict is from Python 2.7. Questions are: * How do you use it in pykickstart and is it possible to change this for backward compatibility? * In general what should we do with this kind of situation where a tool needs the most recent package (in this case pykcisktart) but runs on a stable system like RHEL 6 which has older software stack and the most recent versions are incompatible? -- Alex From clumens at redhat.com Tue Nov 12 18:00:13 2013 From: clumens at redhat.com (Chris Lumens) Date: Tue, 12 Nov 2013 13:00:13 -0500 Subject: pykickstart usage of OrderedDict and SNAKE In-Reply-To: <52823FB0.80007@redhat.com> References: <52823FB0.80007@redhat.com> Message-ID: <20131112180013.GM2175@dhcp55.install.bos.redhat.com> > * How do you use it in pykickstart and is it possible to change this > for backward compatibility? It's not used for very much. See b455a512ccf4bbb84e1ab0ba66c44b59537aa1d3. We could probably come up with another method to accomplish the same thing. If the test cases are written well enough, this should not be a problem. > * In general what should we do with this kind of situation where a > tool needs the most recent package (in this case pykcisktart) but > runs on a stable system like RHEL 6 which has older software stack > and the most recent versions are incompatible? I don't really have a hard rule for these situations. In general, I try to keep pykickstart using an older set of python features and just correct issues as they are brought to my attention. I do not test master on RHEL6, but that could also be done. - Chris From clumens at redhat.com Tue Nov 12 21:56:24 2013 From: clumens at redhat.com (Chris Lumens) Date: Tue, 12 Nov 2013 16:56:24 -0500 Subject: pykickstart usage of OrderedDict and SNAKE In-Reply-To: <20131112180013.GM2175@dhcp55.install.bos.redhat.com> References: <52823FB0.80007@redhat.com> <20131112180013.GM2175@dhcp55.install.bos.redhat.com> Message-ID: <20131112215624.GE5468@dhcp55.install.bos.redhat.com> > It's not used for very much. See > b455a512ccf4bbb84e1ab0ba66c44b59537aa1d3. We could probably come up > with another method to accomplish the same thing. If the test cases are > written well enough, this should not be a problem. See: https://lists.fedorahosted.org/pipermail/anaconda-patches/2013-November/007158.html - Chris From atodorov at redhat.com Thu Nov 14 12:36:32 2013 From: atodorov at redhat.com (Alexander Todorov) Date: Thu, 14 Nov 2013 14:36:32 +0200 Subject: pykickstart usage of OrderedDict and SNAKE In-Reply-To: <20131112215624.GE5468@dhcp55.install.bos.redhat.com> References: <52823FB0.80007@redhat.com> <20131112180013.GM2175@dhcp55.install.bos.redhat.com> <20131112215624.GE5468@dhcp55.install.bos.redhat.com> Message-ID: <5284C3D0.50801@redhat.com> ?? 12.11.2013 23:56, Chris Lumens ??????: >> It's not used for very much. See >> b455a512ccf4bbb84e1ab0ba66c44b59537aa1d3. We could probably come up >> with another method to accomplish the same thing. If the test cases are >> written well enough, this should not be a problem. > > See: > > https://lists.fedorahosted.org/pipermail/anaconda-patches/2013-November/007158.html > Thanks Chris, can you rebuild the package in Koji? -- Alex From clumens at redhat.com Thu Nov 14 14:22:56 2013 From: clumens at redhat.com (Chris Lumens) Date: Thu, 14 Nov 2013 09:22:56 -0500 Subject: pykickstart usage of OrderedDict and SNAKE In-Reply-To: <5284C3D0.50801@redhat.com> References: <52823FB0.80007@redhat.com> <20131112180013.GM2175@dhcp55.install.bos.redhat.com> <20131112215624.GE5468@dhcp55.install.bos.redhat.com> <5284C3D0.50801@redhat.com> Message-ID: <20131114142255.GK5468@dhcp55.install.bos.redhat.com> > Thanks Chris, > can you rebuild the package in Koji? Yep, I was waiting for the recent network changes to go in which just happened. - Chris From atodorov at redhat.com Thu Nov 14 15:15:24 2013 From: atodorov at redhat.com (Alexander Todorov) Date: Thu, 14 Nov 2013 17:15:24 +0200 Subject: pykickstart usage of OrderedDict and SNAKE In-Reply-To: <20131114142255.GK5468@dhcp55.install.bos.redhat.com> References: <52823FB0.80007@redhat.com> <20131112180013.GM2175@dhcp55.install.bos.redhat.com> <20131112215624.GE5468@dhcp55.install.bos.redhat.com> <5284C3D0.50801@redhat.com> <20131114142255.GK5468@dhcp55.install.bos.redhat.com> Message-ID: <5284E90C.60504@redhat.com> ?? 14.11.2013 16:22, Chris Lumens ??????: >> Thanks Chris, >> can you rebuild the package in Koji? > > Yep, I was waiting for the recent network changes to go in which just > happened. > Thanks, I can confirm this fixes my issue.